Assessment of yield stability in barley using univariate and multivariate statistical models

Investigating genetic variability using the phenotypic performance of genotypes is fundamental in a breeding program. Therefore, an assessment of yield performance and yield stability is necessary for yield trials performed in different environments to identify high-yield potential and stable cultivars. In this study, we used 17 univariate and 15 multivariate stability models to investigate the effects of genotype (G), environment (E), and G × E interaction (GEI) on the yield performance of 32 barley genotypes evaluated in 10 environments (locations and years). The main effects were significant (P < 0.01) and accounted for 86.6%, 2.22%, and 11.73% of genotypes, environments, and GEIs of the total variation, respectively. GGE biplot ‘which-won-where’ polygon, divided the environments into five groups, and the genotypes into six groups, among eight genotypes with mean grain yield (GY) superior to the overall mean (4.43 tons ha ^− 1 ). The Spearman's correlation analysis indicated that GY (tons ha ^− 1 ) was significantly and positively correlated (P < 0.01) with Tai’s stability statistics (b _i and α _i ), Perkins and Jinks’s stability parameters ( B _i ), and Roemer’s environmental variance (S _xi ² ) as univariate stability measures. Furthermore, GY had a positive correlation with Ketata’s plotting mean rank (δ gy), Thennarasu’s nonparametric measures (NPI ⁽³⁾ and NPI ⁽⁴⁾ ), Nassar and Huhn’s nonparametric measures (S ^{I 6} and S ^{I 3} ), Fox’s TOP-rank stability (TOP), and the yield stability index (YSI) as multivariate measures of stability. The univariate and multivariate stability models showed that genotypes G32, G1, and G27 were the most stable genotypes with minimal yield variation across environments. Furthermore, G13, followed by G14, G15, and G23 were the most stable genotypes based on multivariate measures only. Accordingly, it might be safe to utilize the stability parameters of different groups with respect to static and dynamic concepts of stability to avoid the possibility of estimating the same concept of stability. Therefore, for the evaluation of genotype stability, a combination of univariate and multivariate stability models is recommended for the selection of “ideal genotypes” for high-yield potential and stable cultivars.